Emerging data fusion methodology in multisensor integrated navigation for challenging environments

随着低成本传感器(MEMS IMU、相机、激光雷达等)的日益普及，挑战是如何将它们最佳地集成在一起，以在具有挑战性的环境中获得准确可靠的导航解决方案，例如GNSS退化或被拒绝的环境。与此同时，对高质量多传感器组合导航的需求不断增长，推动着越来越多的研究和开发活动和应用。因此，这项研究的长期目标是在具有挑战性的环境中发展新兴的多传感器组合导航的数据融合方法。其重点是考虑到快速的技术进步和经济生产率，创新多传感器集成战略，提高系统精度和可靠性，以满足日益增长的无缝室内和室外导航需求，并探索新的应用。 长期研究目标将通过实现拟议的短期目标来实现。首先，在上一次发现资助计划期间开发的通用多传感器集成策略，通过使用运动平台的运动学模型作为系统模型的核心，使传感器测量可以直接参与卡尔曼滤波测量更新，而不区分核心IMU和其他辅助传感器。这对于低成本传感器的使用至关重要，例如MEMS IMU。然而，为了其实用性，这种通用的多传感器集成策略将通过有效减少高速率测量更新、最佳状态选择、可选运动学模型之间的平滑转换、系统级的解评估和传感器级的误差分析等理论和实践方面的改进和优化而得到进一步改进和优化。其次，将通过对低成本IMU阵列进行单独建模和开发低成本IMU自动校准方法来调查低成本IMU阵列的潜力。第三，我们将开发LiDAR-视觉里程计，以进一步增强多传感器(GNSS接收器/IMU/相机)组合导航，以便在具有挑战性的环境中提供更好的导航解决方案，并进一步建立一个互补集成的轨迹确定和3D测绘解决方案的框架。 该计划将极大地促进它们的进步，因为它具有实用的通用多传感器集成战略的意义，IMU阵列建模的新方法和低成本的IMU自动校准过程，以及弹道确定和3D测绘的互补集成解决方案的框架。这一研究项目的成功将极大地影响到应对由于低成本传感器带来的好处以及在具有挑战性的工作环境中的复杂性而带来的巨大挑战，并在一定程度上改善这一领域缺乏高要求的高素质人才的状况

With the increasing availability of low-cost sensors (MEMS IMUs, cameras, LiDARs etc.), the challenge is how to optimally integrate them to reach accurate and reliable navigation solution in challenging environments, e.g., GNSS degraded or denied environments. Meanwhile, the increasing demand for high quality multisensor integrated navigation is continually driving more and more research and development activities and applications. Hence, the long-term goal of this research is to develop the emerging data fusion methodology in multisensor integrated navigation for challenging environments. Its focus is placed on innovating multisensor integration strategies in consideration of rapid technological progress and economic productivity, enhancing system accuracy and reliability for the growing needs of seamless indoor and outdoor navigation, and exploring new applications. The long-term research goal will be achieved through fulfilling the proposed short-term objectives. First, the general multisensor integration strategy, developed during the last Discover Grant Program, has made possible for the sensor measurements to directly participate in Kalman filter measurement updates without distinguishing between the core IMU and other aiding sensors by employing a kinematic model of a moving platform as the core of the system model. This is paramount to the utilization of low-cost sensors, e.g., MEMS IMUS. However, for its practicability, this generic multisensor integration strategy will further be improved and optimized through theoretical and practical enhancements such as effective reduction of high rate measurement updates, optimal state selections, smooth transitions between alternative kinematic models, solution evaluation at system level, and error analysis at sensor level. Second, the investigation of the potential of low-cost IMU arrays will be conducted by modeling them individually and developing low-cost IMU auto calibration methodology. Third, we will develop LiDAR-visual odometry to further enhance multisensor (GNSS receivers/IMUs/Cameras) integrated navigation for the better navigation solution in challenging environments, and further a framework for a complementarily-integrated solution of trajectory determination and 3D mapping. This program will greatly contribute to their advancement by its significance of a practically applicable generic multisensor integration strategy, the new approach to model IMU arrays and the low-cost IMU auto calibration process and a framework for a complementarily-integrated solution of trajectory determination and 3D mapping. The success of this research program will greatly impact on meeting the great challenges due to the benefits of low cost sensors and the complication in challenging working environments and improving the lack of highly demanded high-qualified personnel in this area to some extent